Note

Notebook name: 02a__Intro-Swarm-viresclient.ipynb (download .ipynb)
Alternative view with nbviewer - sometimes the formatting below can be messed up as it is processed by nbsphinx

# Introduction to Swarm access through VirES¶

Authors: Ashley Smith

Abstract: VirES is a server/client architecture to help access time series of Swarm data and models. Access is enabled through a token generated on the website, and a Python package, viresclient, which provides the connection with the Python ecosystem (e.g. xarray).

[1]:

# %load_ext watermark
# %watermark -i -v -p viresclient,pandas,xarray,matplotlib


VirES (Virtual environments for Earth Scientists) is a platform for data & model access, analysis, and visualisation for ESA’s magnetic mission, Swarm

This tutorial introduces the Python interface to VirES, viresclient. We demonstrate usage of the primary Swarm magnetic dataset (SW_OPER_MAGA_LR_1B) and geomagnetic field models produced as part of the Swarm mission. Some knowledge of pandas and matplotlib is assumed.

Run this on the VRE, where viresclient is already installed, or check the instructions to set it up on your own Python environment.

## 0. Configuration¶

To access VirES through Python code, you first need to generate an access token from your VirES account:

1. Create a user account at https://vires.services/ if you haven’t already done so
2. Login there and go to settings (top right: the button with your user name)
3. Select “Manage access tokens” and follow the instructions to create a token (see first part of video above)
4. Copy the token into the box which will appear when you run the code below
5. The default has now been configured and you won’t need to provide the token again.

Run this code and you will be prompted to enter the access token. The token and site url are stored in a configuration file at ~/.viresclient.ini. You may generate and set a new token, or revoke old tokens, at any point.

NB: You will need to repeat this procedure if you are running viresclient from different machines / user spaces.

[2]:

from viresclient import set_token

set_token("https://vires.services/ows", set_default=True)
# (user is now prompted to enter the token)

Generate a token at https://vires.services/accounts/tokens/
Enter token: ································

Token saved for https://vires.services/ows


## 1. Fetching some data¶

Import the SwarmRequest object which provides the VirES interface, and datetime which gives convenient time objects which can be used by SwarmRequest.get_between()

[3]:

from viresclient import SwarmRequest
import datetime as dt


The following code will fetch one day (i.e. around 15 orbits) of the scalar (F) measurements from Swarm Alpha.

Several keyword arguments can be provided to .set_products() to specify the type of data you want. The measurements available depend on the collection chosen in .set_collection(). The same set of auxiliaries are available for all collections - here we also choose to fetch the MLT - magnetic local time.

sampling_step="PT10S" downsamples the data to 10 seconds, from the MAGx_LR default of 1 second. If no sampling_step is provided, the data will be accessed in its original form (i.e. here, 1-second sampling). These strings to choose the sampling_step should be provided as ISO 8601 durations (e.g. "PT1M" for 1-minute sampling).

start_time and end_time in .get_between() together provide the time window you want to fetch data for - executing this line causes the request to be processed on the server and the data returned to you. NB: this returns data up to but not including end_time. Alternatively we can provide the start and end times as ISO_8601 strings.

[4]:

# Set up connection with server
request = SwarmRequest()
# Set collection to use
request.set_collection("SW_OPER_MAGA_LR_1B")
# Set mix of products to fetch:
#   measurements (variables from the given collection)
#   models (magnetic model predictions at spacecraft sampling points)
#   auxiliaries (variables available with any collection)
# Also set additional configuration such as:
#   sampling_step
request.set_products(
measurements=["F"],
sampling_step="PT10S",
auxiliaries=["MLT"]
)
# Fetch data from a given time interval
data = request.get_between(
start_time=dt.datetime(2016,1,1),
end_time=dt.datetime(2016,1,2)
)

[1/1] Processing:  100%|██████████|  [ Elapsed: 00:01, Remaining: 00:00 ]


The data is now contained within the object which we called data:

[5]:

data

[5]:

<viresclient._data_handling.ReturnedData at 0x7fd88588e6d8>


The data is temporarily stored on disk and not yet loaded into memory - the ReturnedData object is actually a wrapper around a temporary CDF file which could be written to disk directly:

[6]:

data.to_file("test_file.cdf", overwrite=True)

Data written to test_file.cdf


… but it is possible to directly transfer it to a pandas.DataFrame object:

[7]:

df = data.as_dataframe()
print(type(df))

<class 'pandas.core.frame.DataFrame'>

[7]:

Longitude F Spacecraft Radius Latitude MLT
2016-01-01 00:00:00 92.793967 46935.8083 A 6833853.08 -72.499224 1.732929
2016-01-01 00:00:10 93.091639 46908.3246 A 6833864.74 -73.130685 1.506584
2016-01-01 00:00:20 93.414902 46878.3804 A 6833875.98 -73.761537 1.263233
2016-01-01 00:00:30 93.766833 46846.0555 A 6833886.81 -74.391708 1.002926
2016-01-01 00:00:40 94.151014 46811.5549 A 6833897.24 -75.021114 0.726345

… or a xarray.Dataset:

[8]:

ds = data.as_xarray()
print(type(ds))
ds

<class 'xarray.core.dataset.Dataset'>

[8]:

<xarray.Dataset>
Dimensions:     (Timestamp: 8640)
Coordinates:
* Timestamp   (Timestamp) datetime64[ns] 2016-01-01 ... 2016-01-01T23:59:50
Data variables:
Spacecraft  (Timestamp) object 'A' 'A' 'A' 'A' 'A' ... 'A' 'A' 'A' 'A' 'A'
Latitude    (Timestamp) float64 -72.5 -73.13 -73.76 ... 29.82 30.46 31.1
MLT         (Timestamp) float64 1.733 1.507 1.263 ... 17.16 17.16 17.16
Longitude   (Timestamp) float64 92.79 93.09 93.41 ... -95.37 -95.37 -95.37
F           (Timestamp) float64 4.694e+04 4.691e+04 ... 3.83e+04 3.861e+04
Radius      (Timestamp) float64 6.834e+06 6.834e+06 ... 6.823e+06 6.823e+06
Attributes:
Sources:         ['SW_OPER_MAGA_LR_1B_20160101T000000_20160101T235959_050...
MagneticModels:  []
RangeFilters:    []

Try plotting some things to visualise the data. The following shows the variation in field strength measured by the satellite as it passes between high and low latitudes, varying from one orbit to the next as it samples a different longitude.

[9]:

df.plot(y="F")
df.plot(y="F", x="Latitude")
df.plot(y="Latitude", x="Longitude")
df.plot(y="Latitude", x="Longitude", c="F", kind="scatter");


## 2. Fetching model evaluations at the same time¶

Various (spherical harmonic) models of the geomagnetic field are produced as Swarm mission products and these are available through VirES. They are evaluated on demand at the same points and times as the data sample points. Here we ask for the MCO_SHA_2D model, a dedicated core field model produced from Swarm data. By supplying residuals=True we will get the data-model residuals, named in the dataframe as F_res_MCO_SHA_2D.

[10]:

request = SwarmRequest()
request.set_collection("SW_OPER_MAGA_LR_1B")
request.set_products(
measurements=["F"],
models=["MCO_SHA_2D"],
residuals=True,
sampling_step="PT10S"
)

data = request.get_between(
start_time=dt.datetime(2016,1,1),
end_time=dt.datetime(2016,1,2)
)

df = data.as_dataframe()

[1/1] Processing:  100%|██████████|  [ Elapsed: 00:01, Remaining: 00:00 ]

[10]:

2016-01-01 00:00:00 92.793967 111.980426 A 6833853.08 -72.499224
2016-01-01 00:00:10 93.091639 108.278648 A 6833864.74 -73.130685
2016-01-01 00:00:20 93.414902 104.358770 A 6833875.98 -73.761537
2016-01-01 00:00:30 93.766833 100.296102 A 6833886.81 -74.391708
2016-01-01 00:00:40 94.151014 96.294332 A 6833897.24 -75.021114

The core field has been removed from the data so the amplitudes are much smaller. Can you interpret the new signals in terms of external fields, i.e. from the ionosphere and magnetosphere?

[11]:

df.plot(y="F_res_MCO_SHA_2D")
df.plot(y="F_res_MCO_SHA_2D", x="Latitude")
df.plot(y="Latitude", x="Longitude", c="F_res_MCO_SHA_2D", kind="scatter");


## 3. More complex model handling¶

We can also remove a magnetospheric field model at the same time, by specifying a new model (which we call MCO_MMA here, but can be named whatever you like) which is the sum of core and magnetospheric models.

See smithara/viresclient_examples/model_details.ipynb for more examples of this - it is also possible to specify the spherical harmonic degrees (min/max) to use, and to provide your own .shc model.

The remaining signal is now primarily due to the ionosphere.

Note that here I am instead using the CI (comprehensive inversion) core and magnetosphere models (2C).

[12]:

request = SwarmRequest()
request.set_collection("SW_OPER_MAGA_LR_1B")
request.set_products(
measurements=["F"],
models=["MCO_MMA = 'MCO_SHA_2C' + 'MMA_SHA_2C-Primary' + 'MMA_SHA_2C-Secondary'"],
residuals=True,
sampling_step="PT10S",
auxiliaries=["MLT"]
)

data = request.get_between(
start_time=dt.datetime(2016,1,1),
end_time=dt.datetime(2016,1,2)
)

df = data.as_dataframe()

[1/1] Processing:  100%|██████████|  [ Elapsed: 00:03, Remaining: 00:00 ]

[12]:

Longitude F_res_MCO_MMA Spacecraft Radius Latitude MLT
2016-01-01 00:00:00 92.793967 81.105345 A 6833853.08 -72.499224 1.732929
2016-01-01 00:00:10 93.091639 76.994628 A 6833864.74 -73.130685 1.506584
2016-01-01 00:00:20 93.414902 72.672585 A 6833875.98 -73.761537 1.263233
2016-01-01 00:00:30 93.766833 68.209912 A 6833886.81 -74.391708 1.002926
2016-01-01 00:00:40 94.151014 63.805549 A 6833897.24 -75.021114 0.726345
[13]:

df.plot(y="F_res_MCO_MMA")
df.plot(y="F_res_MCO_MMA", x="Latitude")
df.plot(y="Latitude", x="Longitude", c="F_res_MCO_MMA", kind="scatter");